Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I wish I followed this advice when I was setting up lambdas on aws. There was this general distaste for "java" in the dev community. Doing a new project and this time around I am using java, and it works so beautifully. The second choice would have been python, but I like the strict type checking.


Do you find that using the JVM in Lambda affects your cold start times? If so, do you have a pre-warm strategy and how does it work? Or maybe it doesn't matter for your application?


Try Micronaut, which is a micro framework specifically designed for superfast startup time and low memory footprint. It also works with GraalVM.


I did when I used Java Lambdas behind API gateway for a largish public API. I ended just setting the HTTP purification tests to run every minute. Solved the issue and let me know how stable it was. Not a huge issue in the end.


Lambda does re-use the JVM instances where it can to alleviate that.


That only matters if your concurrency is one. If you get 10 requests each before the previous one completes you will get cold starts every other call or so.. its manageable, just takes effort.


That is hilariously ironic considering 90% of AWS is written in Java.


Have you got a source or more information on this? I'm not disputing the statement - but I've seen several different languages claimed to run "most of AWS" (including Haskell at one point) and it'd be good to have a firmer answer.


I used to work for AWS. The API framework that powers most if not all of AWS API's is called "Coral" and is Java based.


It's like people think Java only supports concurrency via thread pool and isn't capable of event loop / async io models.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: